Surface Defect 3D Reconstruction of Object Based on Binocular Vision

Discussed the basic principle, key steps and main framework of 3d reconstruction system based on binocular vision from whole aspects. To avoid the problem of repeated or error feature matching and time-consuming in traditional matching rules based on sift algorithm and Euclidean distance. This paper presents an improved sift algorithm using cosine distance to match feature in a limited range. Reconstructing and texture mapping the 3d information of surface defect of object by the software of opengl. The experimental results show this method can enhance the speed and the precision of feature matching and be more precise to the surface defect 3d reconstruction of object.